dpa Launches dpa-iq to Provide Trusted Information for AI Agents

The German press agency dpa (Deutsche Presse-Agentur) is launching a new platform called dpa-iq, designed to serve as a trusted information layer for AI-powered agentic systems. The initiative marks a strategic pivot for the agency as it adapts its distribution model to a landscape where AI agents increasingly perform the knowledge and information work previously handled by journalists and editors.

Founded in 1949, dpa operates as a joint venture comprising approximately 170 German media companies. For 77 years, the agency has focused on two primary areas: providing global news events through its wire service and offering a news hub platform for topic-based research. However, Yannick Franke, the AI Team Lead at dpa, stated during the WAN-IFRA Frankfurt AI Forum that this established framework is being upended by the rise of artificial intelligence.

Franke noted that the industry is undergoing a fundamental shift because a lot of knowledge work and information work is now done not directly by editors and journalists, but by intermediates – by AI systems.

This transition has led dpa to address existential questions regarding how to fulfill its mission of providing trusted information in a world where AI agents handle the bulk of information-seeking. The agency’s response is based on the core premise that dpa should provide easy access to trusted information wherever its customers are conducting AI-based work.

An API-Driven Layer for AI Agents

Currently in private preview, dpa-iq is designed as an API product that replicates the functionality of the agency’s news hub but specifically for AI systems. While journalists use the news hub to seek reliable information on specific topics, dpa-iq allows AI agents to query the system to retrieve verified data necessary to complete their assigned tasks.

For example, a journalist could instruct an AI agent to find specific materials, such as recent news regarding the situation in Iran, images of a politician in a specific location, or relevant B-roll video content. The agent would then access dpa-iq to retrieve these verified items.

Initially, the platform will exclusively feature dpa content, including articles, videos, images, audio files, and databases. Dpa is also in discussions with partners to integrate additional sources into the system.

One such area of expansion is sports data. Franke explained that users of the news hub frequently seek sports information, but these are often data-driven questions rather than requests for journalistic articles, noting that we don’t have a dpa article for every data question. dpa will integrate information from a data provider that structures and collects data from various German government bodies across different geographical levels.

Modular Architecture and Technical Integration

The architecture of dpa-iq is built as a multi-source retrieval system rather than a single monolithic data pillar. This modular design allows the agency to integrate new data sources in the future and swap out different vendors, technologies, or services as the AI field evolves.

The platform includes a multi-source retrieval endpoint that allows customers to ask questions and search across various sources for relevant information. It also features a generation endpoint capable of generating answers to questions. Franke clarified that the goal is not to create a chatbot, stating that such a feature is intended for commodity purposes to make it easier to deploy applications on top of the platform.

To support product teams, dpa-iq functions as an API management platform with granular controls. This includes the ability to define rate limits and access rights for individual users in multi-user environments.

Ecosystem and Automation

To ensure its content is available where clients are performing AI work, dpa is developing integrations for platforms such as OpenAI and the AI integration platform Langdock. The agency is also enabling compatibility with process automation tools, including Make, n8n, and Zapier.

As a demonstration of these capabilities, the dpa team built a workflow using these automation tools. In the demo, the system scans the dpa-iq archive every morning at 6:00 am for specific information and organizes the materials into a newsletter that is ready for publication.